Abstract
Aerial video stabilization system aims to remove undesired motion in aerial video. This motion is the result of undesired movement of mobile sensor. In this article we present a new video stabilization system for Unmanned Aerial Vehicles (UAV). Our system is based on keypoints tracking. We use Scale Invariant Feature Transform (SIFT) keypoint detection, and matching to estimate parameters of affine transformation model. Then, Kalman filter with median filter is applied to remove video noise. A number of real aerials videos surveillances demonstrate that this method can achieve good performance. (C) 2013 The Authors. Published by Elsevier B.V.